Issue 26 August 2010

Flotation circuit optimisation using

modelling and simulation software
Authors: Rob Coleman & Pertti Lamberg
How is your flotation plant running? Is it achieving target throughput,
recovery and grade? If not, why not? Could it be due to feed ore
Outotec Australias
quarterly e-newsletter
Contents
variability, grinding and liberation issues or reagent type and dosage?
Or could it be insufficient residence time since you are now operating
at 50% more throughput than originally designed?
When was the mineralogy of the feed last measured? Or a
metallurgical survey to determine accurate flows around the circuit
conducted? How about the gas dispersion in each flotation cell are
bubbles of the correct distribution for given particle sizes being
d?
1
Contents
Flotation circuit
optimisation: 1
Protecting your mill: 5
Angas Mine: 9
created?
It is nearly impossible to optimise your flotation circuit performance
without knowing the answers to these questions.
Over the past decade there has been a significant advance in flotation
circuit optimisation through performance benchmarking using
metallurgical modelling and steady-state computer simulation. This
benchmarking incl des traditional meas res s ch as grade and
Pulp level
control in
flotation circuits
benchmarking includes traditional measures, such as grade and
recovery, as well as new flotation measures, such as ore floatability,
bubble surface area flux and froth recovery.
Such circuit optimisation is a powerful tool in achieving the best
possible flotation performance. Outotecs HSC Chemistry software,
which incorporates a new simulation module, can assist with this
optimisation.
Firstly, how do you benchmark your flotation circuit?
There are several ways to do this, including:
Mineralogical assessment of flotation streams: determine the
ore mineralogy, degree of liberation, mineral associations and
locking at different particle sizes.
Metallurgical assessment of performance: perform plant
surveys to determine grades and recoveries of streams around surveys to determine grades and recoveries of streams around
the circuit (requires mass balancing of experimental data). Froth
carry rates, concentrate lip loadings and flotation cell residence
times can also be determined.
Comparison of plant to laboratory performance: determine
the maximum attainable recovery from the ore and compare this
to current plant recovery.
Gas dispersion in each flotation cell: determine the bubble
size distribution air hold-up superficial gas velocity and bubble size distribution, air hold up, superficial gas velocity and bubble
surface area flux.
Froth phase performance: determine the recovery across the
froth phase, froth stability and froth transport distances.
Editor:
Laura White
laura.white@outotec.com
www.outotec.com
Modelling and simulating flotation performance
Output Australia | August 2010 | Page 2
g g p
The results from these studies can then be used to calibrate a floatability component model of the
circuit. The flotation model developed by the AMIRA P9 Project, of which Outotec is a sponsor, is
regarded by industry as the most suitable flotation model to use for circuit optimisation. This
model incorporates ore floatability with flotation cell pulp and froth parameters, residence time,
entrainment and water recovery to the concentrate.
Once the model is calibrated, it can be set-up in a flotation circuit simulator, such as Outotecs
HSC Sim 7.0. The simulator is then able to predict the performance of the flotation circuit under
various hypothetical changes to the operation of the circuit. This can include changes to feed
properties (such as throughput, mineral content and grind size), flotation cell operating properties
(such as air rate and froth depth) and circuit configuration (for example recirculating the cleaner
scavenger tails to the scavengers, or the addition of more flotation cells to increase residence
time).
This provides an extremely powerful tool to assist with optimising flotation performance. The p y p p g p
latest version of HSC Chemistry software, 7.0, includes further optimisations of existing tools such
as a steady state process simulator and flowsheet capabilities, along with further additions.
So, what can HSC Sim do?
Stage 1 flowsheet design
The first stage of any flotation optimisation process is drawing the flowsheet, which is done
graphically by the user. HSC also includes check for error tools to ensure various streams are g p y y
properly connected to the units and the process has input and output streams. If your process is
large, the flowsheet can be split into several pages, ie comminution on one page, flotation on the
next etc, followed by a summary page.
Stage 2 & 3 mass balancing and calibration
The next stage is mass balancing the experimental data for the development of the model. HSC
Sim has a new experimental mode, which can collate, organise and visualise survey or laboratory
data A Mass Balancing and Data Reconciliation module is included and has the following data. A Mass Balancing and Data Reconciliation module is included and has the following
features (Figure 1):
Individual sampling error for each stream and general or individual error model for each
measurement
1D (unsized), 1.5D (sized but no assays) and 2D (size-by-size assays) mass balancing
Various regression options such as least-squares regression
Versatile visualisation tools such as parity charts and stream tables and mass balance
reports p
Once the experimental data has been mass balanced, the model parameters can be determined.
Calibration comes next and includes elements such as global mineralogy and feed streams
(grades and flowrates). The simulator element of optimisation is then ready to run.
Stage 4 - simulation
Most of the more advanced software nowadays includes a mineralogical database. HSC Chemistry
7 0 has over 4 500 different species and more than 13 000 different minerals In simulation you 7.0, has over 4,500 different species and more than 13,000 different minerals. In simulation you
can select the best matching mineral or add your own minerals into the database. It is possible to
set-up HSC so that each person at a site uses and shares the very same database on the local
network. HSC even has a versatile tool for automatically converting elemental assays to mineral
grades.
As mineral processes do not treat minerals, but particles of different sizes and different
Output Australia | August 2010 | Page 3
compositions, it is important the software is designed on that basis. With HSC, you select 5 different
minerals in 5 size fractions, with 3 different behaviour types for each mineral, and HSC will create 75
particles (mineral x size x types). Particles have global properties like size, specific gravity and
composition and each unit uses these particle properties to determine what to do with each particle.
A structure based on particles allows you to load your liberation data from an MLA (Mineral
Liberation Analyser) into the simulator and simulate the process with true (measured) particles. In
the highest level, i.e. with true particles, you can have very detailed information on your process g , p , y y y p
losses and impurities. At particle level you can simulate scenarios like:
How will the change in grind influence the metallurgical performance of the plant?
How will change in liberation influence the metallurgical performance of the plant?
How will the concentrate quality change if we target to reject/accept some of the minerals?
Figure 1 - Mass balance of unsized (1D) or size-by-size (2D), using HSC Sim 7.0
Case study Case study
HSC Sim was used to design the flowsheet of the Esperanza copper-gold deposit in Chile. One
particular task of the software was to understand what benefits, if any, would result from the
inclusion of a SkimAir flash flotation cell in the process cyclone underflows. Using client laboratory
testwork, the resulting elemental data was converted to minerals, and their pilot testwork data was
mass balanced, with models built using both laboratory flotation tests. The simulation was
established on a mineral-by-size by floatability type level. Several different scenarios were
simulated, including: circuit with and without SkimAir; number and type of rougher cells; flash
l i i h d d d t d diff i f d t cleaning; varying head grades and ore types and differing feed rates.
Figure 2 shows the simulation results with and without the inclusion of the SkimAir. The simulations
showed that the SkimAir circuit produces 7.6% higher overall recovery for gold and 2.2% for copper.
SkimAir technology was selected following these results and Esperanza will be commissioned in
2010.
Output Australia | August 2010 | Page 4
Summary Summary
The use of software tools for
the simulation of the flotation
circuit are a major advance in
flotation modelling and
optimisation. Steady-state
simulators can be used for
tasks including circuit
di i b ttl k
SkimAir
Rougher
Grinding
Cyclone
Cycl one UF
SK Conc
Cycl one OF
Rougher Tai l
Rougher Conc
Scavenger Conc
Fi nal Ta
100
502
0
100.0
17.7
82.3
20.1
Au Rec%
0
Esperanza Flotation Circuit Modeling
Outotec Research, Pertti Lamberg
February2008
Porphyry Conventional
HSC Chemistry 6.1
diagnosis, process bottle-neck
identification, ascertaining the
effect of various parameters
on metallurgical performance
and sizing process units
properly.
Some of the more advanced
simulation tools such as
g
Cleaner
Scavenger
Ore
SK Tai l
Cl eaner Conc
Cl eaner Tai l
g
Scavenger Tail
CUF+SKTail
Flash Cleaning
Fl ash Cl eaner Conc
Fl ash Cl eaner Tai l
100
788 0
402
79.9
2.46
402
0
167
86.8
84.3
Final concentrate
Cu Grade % 30.1
simulation tools, such as
Outotecs HSC Sim, enables
you to simulate mineral
processes in different levels,
from comminution circuits
with sizes and no composition,
through to flotation processes
with minerals by size by
floatability components to full
Scavenger
Final Conc
Fl ash Cl eaner Conc
79.9
0
Cu Grade % 30.1
Cu Recovery 94.7
Au gr ade g/t 13.9
Au Recover y 79.9
SkimAir
Rougher
Cyclone
Cycl one UF
Cycl one OF
Fi nal Tai l
163
40.6
Au Rec%
Esperanza Flotation Circuit Modeling
Outotec Research, Pertti Lamberg
February2008
Porpyry Flash
HSC Chemistry 6.1
floatability components, to full
processes with true particles
with measured liberation
(MLA) data. Additionally,
experimental data can be
collected, elemental assays
converted to mineral grades,
circuits mass balanced and
data reconciled in mineral-by-
Grinding
Cleaner
Ore
SK Tail
SK Conc
Rougher Tai l
Rougher Conc
Cl eaner Tai l
Scavenger Conc
Fi nal Tai l
CUF+SKTail
Flash Cleaning
100
320
204
104
12.7
0
27.9
0.24
12.9
104
59.4
55.0
27 3
27.1
data reconciled in mineral by
size level. Powerfully, different
scenarios can be run with the
process simulation, saving
hundreds of man hours (and
potential human inaccuracies)
in the process.
Scavenger
Cl eaner Conc
Scavenger Tail
Final Conc
CUF+SKTail
g
Fl ash Cl eaner Conc
Fl ash Cl eaner Tai l
27.7
87.1
0
0
27.3
Final concent rate
Cu Grade % 17.9
Cu Recovery 96.9
Au grade g/t 8.8
Au Recovery 87.1
Figures 2. Gold recovery (distribution) in the Esperanza flotation circuit
without SkimAir (above) and where cyclone underflow is directed to the
SkimAir flotation machine (below).
_______________________________________________________________________________________________
Dr Rob Coleman is currently Technology Leader Flotation for Outotec Pty Ltd in Australia. He has a
Chemical Engineering degree and a Doctorate in Minerals Processing. Rob has over 15 years
experience in the operation, modelling and optimisation of flotation circuits.
Dr Pertti Lamberg is Senior Technology Adviser at Outotec Research based in Pori, Finland. Pertti has
a Doctorate in Geology, and has over 10 years of experience in modelling and simulating of minerals
processes He is one of the developers of the HSC Chemistry software
If you would like more information, click here to contact
rob.coleman@outotec.com
processes. He is one of the developers of the HSC Chemistry software.